HA-IDS: A heterogeneous anomaly-based intrusion detection system

Abstract

In this research, we propose a heterogeneous anomaly-based intrusion detection system (HA-IDS) which is built on both of Field-Programmable Gate Array (FPGA) and Graphics Processing Unit (GPU) platforms. An essential anomaly-based IDS comprises of the two main components: Feature Construction Module (FC) to extract and collect network header information, and Classification Module (CM) to categorize connections' behaviors. In our proposed system, FC is implemented on FPGA to handle large-scale data while CM is deployed on GPU using Back-propagation Neuron Network to utilize the parallel computing for improving the system performance. We employ our proposed system on GPU — GIGABYTE GeForce GTX 1080 and FPGA — Xilinx Virtex-5 XC5VTX240T chip. The experiment result shows that the training outcome on GPU is faster than that on CPU by up to 12x. The testing system throughput in real-time is 200Mbps with more than 80% in Accuracy Rate. This work proves that this is promising to take advantages of these two platforms to apply to a particular issue.